TAIMWAS: A Trustworthy AI Model-based Intelligent Chronic Wound Analysis System

点击翻译按钮获取中文摘要

The project will develop intelligent software/app for mobile/tablet that can work as an end-to-end solution for a fully automated wound analysis system. This software will deploy multiple **robust** and **verifiable** deep learning (DL) models to analyse 2D/3D images and extract information about various aspects of wounds for intelligent and efficient management. The DL models will be trained/updated through a federated learning (FL) approach to improve data diversity, preserve users' privacy, and minimise security risk. We will also use a deep generative model to verify the accuracy of the models trained. Thus, the robustness and verifiability of the models will make our solution **TAIMWAS** trustworthy. And this is the first **trustworthy AI system** that offers end-to-end wound management services using FL using multi-site and diverse image data. In addition, the solution's accurate non-contact measurements, automated facility reporting, and analytics will enhance clinical productivity, reduce clinical variability and quality patient outcomes, and minimise risk & liability.